DE102015225900B3 - Method and device for camera-based traffic sign recognition in a motor vehicle - Google Patents

Abstract

A method for camera-based traffic sign recognition in a motor vehicle comprises capturing at least one image of a space in front of a vehicle with a camera and performing a pattern recognition to detect traffic signs in the at least one recorded image. After the detection of a traffic sign in a first image, a sequence of further images of the space in front of the vehicle is taken with the camera, which tracks traffic signs recognized in the first image in the sequence of the further images, and a resizing and / or shifting of a position the detected and traced traffic sign in the respective images on the sequence of further images determines a first speed value indicating the speed with which the detected and traced traffic sign is approaching the camera and / or passing by the camera. The first speed value is compared with a second speed value provided by a speedometer of the motor vehicle, which represents a driving speed of the motor vehicle. If the difference in speed values is above a threshold, the detected traffic sign is discarded. Otherwise, the recognized traffic sign is signaled to a driver and / or a driver assistance system of the motor vehicle.

Description

In many motor vehicles, traffic sign recognition is provided by means of a camera which records images of a traffic space lying in front of the motor vehicle in the direction of travel. The images taken by the camera are analyzed by means of a computer and appropriate software, and traffic signs are recognized by pattern recognition in the images. Contents of recognized traffic signs are extracted and the extracted information is displayed to a driver or incorporated into a driver assistance system. In particular, information about speed limits is displayed to give the driver another opportunity to perceive it in the event that the driver has missed an appropriate traffic sign.

The detection of traffic signs by means of a camera and an automatic image analysis has the advantage over a display of traffic signs from a database and corresponding position determination of the vehicle that temporarily established traffic signs, for example in construction sites, can always be considered, while in database-based systems of the update frequency depends on the database.

At the back of some vehicles, especially in trucks and buses, but also on caravans, stickers in the form of a speed limit signaling traffic signs are often attached, which should inform the following traffic about a current maximum speed for this vehicle. Among the traffic signs signaling a speed limit are also miniaturized images of city entrance panels intended to indicate a connection of the vehicle owner with a location. All of these "traffic signs" may be recognized by the traffic sign recognition and at least one resulting traffic recommendation or restriction may be erroneously displayed to a driver or even used by a driver assistance system to automatically affect the driven route or speed. Such misrecognition takes place especially in convoys or overtaking, because then the traffic sign representing sticker frontally or laterally recognized by the motor vehicle, so it can be well detected by the camera and fits in a conventional scheme for installation of traffic signs.

From the DE 10 2007 021 576 A1 a method for traffic sign recognition by means of a vehicle arranged in a camera is known in which a self-motion of the vehicle is used for a plausibility check.

From the US 2015/0178588 A1 For example, a method for object recognition by means of a camera arranged in a vehicle is known in which a relative movement of an object candidate to the camera is used to distinguish an object from a reflection of the object.

From the EP 2 383 679 A1 a method for detecting traffic signs is known by means of a camera arranged in a vehicle, in which a relative movement between a detected traffic sign is compared with the self-motion of the vehicle in order to exclude non-static traffic signs from further detection.

From the EP 1 383 098 B1 a system for the automatic recognition of traffic signs is known in which a speed of the vehicle is used for camera control and in an image processing of the recognition of the traffic signs is taken into account.

Technical problem

It is an object of the present invention to provide an improved method and apparatus for camera-based traffic sign recognition.

Technical solution

This object is achieved by the method specified in claim 1 and the device specified in claim 8. Advantageous embodiments and further developments of the method and the device are specified in the respective dependent claims.

The inventive method and the device according to the invention for improved traffic sign recognition provide, in addition to the known recognition of traffic signs, to use information provided by a speedometer about the current vehicle speed for a plausibility check of the recognition.

The invention makes use of the recognition that valid traffic signs are usually fixed in place or, for example, if they are attached to a vehicle of the road maintenance, which performs a control trip on a road, move only at very low speed. A traffic sign sticker on a vehicle, such as a truck, will move at a significantly higher speed, e.g. B. on highways with the applicable maximum speed of 80 km / h for trucks.

An inventive method for camera-based traffic sign recognition comprises taking at least one image of a vehicle in front of a space with a camera and performing a pattern recognition for detecting traffic signs in the at least one recorded image. According to the invention, after recognition of a traffic sign in a first image and, if appropriate, after successful pattern recognition, a sequence of further images of the space in front of the vehicle is taken with the camera, and the traffic sign recognized in the first image is tracked in the sequence of the further images.

Under pursuit here is the recognition of the same traffic sign and the determination of a respective position of the recognized traffic sign in the respective further images to understand. Here, it may be sufficient to recognize the traffic sign as such without extracting information conveyed by the traffic sign if the information has already been safely extracted at the first detection. By dispensing with the extraction of information, the required computing power can be reduced. A plausibility check as to whether the tracked traffic sign is actually the previously recognized traffic sign can be made, for example, if the traffic sign is not at an expected position in a subsequent image, or has another basic shape and / or color. The plausibility check can also be carried out with different parameters or methods, depending on how many traffic signs in the image taken by the camera are close to each other, especially if they are similar traffic signs. For example, similar traffic signs may have similar basic shapes and / or similar color combinations.

From the change in size and / or displacement of a position of the detected and tracked traffic sign in the respective images on the sequence of further images, a first speed value is determined which indicates the speed at which the detected and traced traffic sign on the camera and / or on the camera passes by. It exploits the fact that objects in the recording area of a camera with an unchanged focal length when approaching are displayed increasingly larger in successive images. The term "consecutive images" here refers to those successive images which are supplied to the image processing for the purpose of traffic sign recognition. In addition, an object located outside a straight line of sight of the camera in a sequence of images as it approaches the object will change its position within the respective images. If a camera supplies more images per unit of time than required for the purpose of traffic sign recognition according to the invention, only one selection of the images actually taken can be used while other images are not used.

Under the generally accepted assumption that the traffic sign does not change its size, therefore, a speed of approach can be determined from the change in size over time in the image series. In addition, a traffic sign will generally not be so mounted on a lane of a road that a vehicle approaches it head-on, unless the lane ends before the traffic sign or turns left or right. Accordingly, a traffic sign changes its position when the motor vehicle approaches in successive images of a camera oriented forward in the direction of travel. The position of the traffic sign is, starting from the first position at which it was detected, and with unchanged direction of travel, move to the sides and / or up or down in the image section. From the change in the position of the detected and tracked traffic sign, with known and unchanged characteristics of the camera and the optics, also a speed can be determined. Of course, the two possibilities of determining a speed value from the traffic signs tracked in the sequence of the images can also be combined in order to achieve a higher accuracy.

The first speed value is then compared with a second speed value provided by a speedometer of the motor vehicle, which represents a driving speed of the motor vehicle. The speedometer works independently of the camera. When a difference between first and second speed values is below a threshold, the traffic sign or information extracted therefrom is signaled to the driver or the driver assistance system. Otherwise, the traffic sign is rejected as unreliable and information extracted therefrom is not signaled or used because the traffic sign was not stationary and was highly likely to be attached to another vehicle. As an alternative to discarding, an indication of an unsafe traffic sign can be made, so that the driver can decide on the basis of the current traffic situation or the environment whether he has possibly overlooked a traffic sign.

In one embodiment of the method according to the invention, the second speed value is provided via a data bus of the vehicle. For example, speed information may be provided via a CAN bus, similar to a speed-dependent volume control for a car radio. The connection to the data bus can be via cable or wireless, z. B. in accordance with IEEE 802.15.1 (Bluetooth).

The second speed value can be determined, for example, from data from one or more speed sensors on a vehicle transmission or at one or more wheels or drive shafts, or can also be determined from data from a satellite navigation system. Combinations of the different sources of data to determine the second speed value are of course also conceivable.

In one embodiment of the method according to the invention, the recording of further images takes place with a fixed frame rate. The term frame rate in this context refers to the number of pictures per unit of time. This makes it possible to make the determination of the first speed value only dependent on the change in size or position of the detected and tracked traffic sign as well as fixed parameters of the camera and the optics. However, in this embodiment, the accuracy of the determination of the first speed value is higher at lower speeds of the motor vehicle than at high speeds, because the change in the size or the position of a recognized and tracked traffic sign between two images is smaller than at high speeds of the motor vehicle. In addition, especially at a rather low frame rate, with a large change in position between two consecutive frames, the tracking of the detected traffic sign at higher speeds may become difficult, or an increase in tracking errors may occur.

The threshold for the difference between the first and second speed values may be selected to account for the lower accuracy of the speed estimate from the camera images versus the speedometer of the vehicle. Thus, for example, a threshold value or tolerance value may be selected such that a slow change in location of traffic signs is considered valid. Such slow changes in location come z. Example, when at the back of a slow-moving control vehicle of a Straßenmeisterei a traffic sign is attached to regulate the speed or to specify a lane.

In one embodiment of the method according to the invention, the threshold value for the difference between the first and the second speed value depends on the second speed value. Thus, for example, with a fixed frame rate, a lower accuracy of the determination of the first velocity value can be taken into account at higher speeds.

In an embodiment of the method according to the invention, the recording of further images takes place with a frame rate dependent on the second speed value. The frame rate may be slower at low speeds than at high speeds. This can ensure that the changes in the size or position of the recognized and tracked traffic sign in two consecutive images remain within a limited range. This can lead to a simplification and acceleration of the recognition of the traffic sign in the next image because the image area to be searched for in the recognition can be narrowed down and a prediction of the expected size and position is simplified due to a simple mathematical relationship. The frame rate may alternatively or additionally be changed depending on other factors described below, for example to minimize the changes in the image content in successive images.

The change of the frame rate can be done by actually adjusting the frame rate in the camera. However, if the camera does not allow adjustment of the frame rate or, for example, needs to operate at a higher frame rate to perform other tasks than would be required for traffic sign recognition, a change in frame rate may also be achieved by using only every nth frame for traffic sign recognition purposes done, so by omitting images in the image processing. The change of the frame rate can be carried out substantially continuously at the speed, or even in discrete stages for speed ranges.

A lower frame rate at low speeds requires fewer resources in image processing, thereby freeing up for other purposes. Thus, at lower speeds released computing power z. B. are used for improved pedestrian or object recognition, if one assumes a higher probability of the presence of pedestrians or for the driver of a motor vehicle relevant objects in traffic areas in which lower speeds are driven.

In one embodiment of the method according to the invention, in addition to the second speed value, a steering angle, that is to say a variable dependent on a position of a steering device of the motor vehicle, is taken into account in the determination of the first speed value. The consideration of the steering angle together with the second speed value makes it possible to predict or determine a change in the position of the recognized and tracked traffic sign in successive images due to the pivoting of the camera firmly connected to the motor vehicle by changing the direction of travel, and in the determination of corresponding to the first speed value. The values or data representing the steering angle can be provided, for example, by corresponding sensors on the steering device and / or by acceleration or gyro sensors for an electronic stability control of the motor vehicle or by an electronic compass. The electronic compass or the acceleration or gyro sensors can also be arranged in a mobile telephone or smartphone temporarily connected to the vehicle, the corresponding data then being provided via a wired or wireless connection. Preferably, the mobile phone or the smartphone is arranged in the motor vehicle, that it performs movements of the body at the same time and largely identical.

Similarly, a change in the inclination of the vehicle in the longitudinal and / or transverse direction, for example in the case of a changing incline or a changing profile of the surface of the road, can be taken into account in successive images. Thus, when driving on a road with bumps that can not be fully compensated by a suspension of the motor vehicle, the body and thus the camera perform an up-and-down movement or a lateral rolling motion, in the sea as "pounding "Or" roles "is known. Since the camera is firmly connected to the body of the motor vehicle, the area covered by the camera will change accordingly. However, a change in position could erroneously result in the determination of a speed other than the actual speed when determining the first speed value. Even with larger distances between the motor vehicle and the traffic sign, even a slight change in the slope of the road can cause the position of a further ahead traffic sign to move upwards in successive images, for example when driving through a subsidence. This change in position could be interpreted as a higher speed than the actual speed of the motor vehicle, because a comparable change in position in consecutive images on a level track correlates with a higher speed. A consideration of the change inclination of the motor vehicle can thus lead to a higher accuracy of the determination of the first speed value. For the determination of the inclination both in the longitudinal and in the transverse direction of the motor vehicle, sensors already present in the motor vehicle can be used on or in the chassis suspension and / or for an electronic stability control. It is also conceivable to use acceleration sensors, gyro sensors or an electronic compass of a vehicle-mounted satellite navigation system. It is conceivable to use corresponding sensors of a temporarily connected to the vehicle mobile phone or smartphone. The sensors can provide appropriate data via a wired or wireless connection. Preferably, the mobile phone or the smartphone is arranged in the motor vehicle, that it performs movements of the body at the same time and largely identical.

In one embodiment of the method according to the invention, the threshold value above which a traffic sign is rejected is changed as a function of the second speed value, the steering angle and / or a change in the inclination of the vehicle in the longitudinal and / or transverse direction. For this purpose, an assignment rule can be provided which describes an association of two or more values or parameters to one another. The assignment rule can be described in terms of formulas, or it can also be implemented as an assignment table. Assignments can be made continuously or in areas. As a result, a determination of the first speed value that varies with the speed of the motor vehicle, a steering angle or dynamic changes in the body position can be taken into account.

In one embodiment of the method according to the invention, a recognized traffic sign is discarded if it could not be detected or tracked in an uninterruptedly consecutive minimum number of further images. The minimum number may be determined as a function of the second speed value, the steering angle and / or a change in the inclination of the vehicle in the longitudinal and / or transverse direction.

In one embodiment of the method according to the invention, the method pauses when the second speed values correspond to a motor vehicle that is not in motion. Here, the realization is used that recording from further images and tracking of traffic signs while the motor vehicle is standing, provides no meaningfully evaluable information for determining a speed. While the process pauses, ie, does not track the detected traffic sign, and does not detect a speed from a series of images, energy can be saved and / or resources, such as a computer set up to perform the process, can be used by others at this time Purposes are used.

In one embodiment of the method according to the invention, the method pauses when a traffic sign with a speed limit has been detected to very low values, for example 6 km / h. Stickers with such speed limits are sometimes attached to agricultural towing vehicles, but are otherwise very rare.

In one embodiment of the method according to the invention, a database query takes place in order to determine permanently installed traffic signs known at the current position of the vehicle. The database query can be used to further verify the recognition. If the database query does not match, it is also possible to send the result of the discovery to the database. If a traffic sign recognized at one position is not reported to the database by several road users independently of each other and over a certain period of time, the database can be updated accordingly.

In one embodiment of the method, information of a lane recognition is used to improve the traffic sign recognition. Thus, for example, when the motor vehicle travels on a middle lane of a three-lane roadway, a traffic sign recognized on a lane immediately next to it can be discarded, because in this case only one traffic sign located above the lane can be valid.

In one embodiment of the method, information of an assistance system is used which can identify vehicles in front to exclude image areas from the traffic sign recognition that have been identified as belonging to a preceding vehicle.

In the following the invention will be described with reference to the drawing. In the drawing shows

1 a schematic representation of a camera view of a first traffic situation in front of a vehicle at a first time;

2 a schematic representation of the camera view of the first traffic situation in front of the vehicle at a second time;

3 a schematic representation of a camera view of a second traffic situation in front of a vehicle at a first time;

4 a schematic representation of the camera view of the second traffic situation in front of the vehicle at a second time;

5 a schematic representation of a camera view of a third traffic situation in front of a vehicle at a first time;

6 a schematic representation of the camera view of the third traffic situation in front of the vehicle at a second time;

7 an exemplary flowchart of an embodiment of the method according to the invention; and

8th a schematic block diagram of an apparatus for performing the method.

In the figures, the same or similar elements are identified by the same reference numerals.

1 shows a first schematic representation of a camera view of a first traffic situation 100 in front of a vehicle (not shown) at a first time. The traffic situation shows a straight to the horizon 101 running two-lane road 102 , The vehicle, not shown in the figure drives on the right lane in the direction of the horizon 101 , At some distance from the current position of the vehicle is a traffic sign 103 , which signals a speed limit of 80 km / h. The camera, which is not shown in the figure, provides images of the traffic situation to a traffic sign recognition module (also not shown in the figure), which may include dedicated software that is executed on suitable hardware. Methods for traffic sign recognition are known from the prior art. According to the invention, when a traffic sign has been detected, in this example the traffic sign 103 tracking traffic signs in subsequently captured images. For this purpose, an area occupied by the traffic sign or a lying around the traffic sign area of the image can be marked, which must be recognized in the subsequently recorded images again. The of the Camera captured scene changes due to the movement of the camera with the vehicle. This change is reflected, for example, in a change in size of objects, but also in a change of location of objects within the image taken by the camera. From a known time interval between two images of the camera and a change in size and / or change of location of objects, a speed of the camera and thus of the vehicle can be determined. Immutable properties of a lens of the camera, such as nonlinearities or distortions in certain image areas can be taken into account accordingly.

2 shows a schematic representation of the camera view on the first traffic situation 100 in front of the vehicle at a second time. The vehicle has moved forward a certain distance and is the traffic sign 103 come closer. Accordingly, the traffic sign 103 in the picture now larger and is located at a different position in the picture. The change of position compared to in 1 The camera view shown at the first time is indicated by the arrows 205 and 206 clarified. From the time difference between the first and the second time as well as the change of the position of the traffic sign 103 According to the invention, a speed is determined with which the camera has moved towards the traffic sign. The determined speed is compared with a speed value provided by a speedometer of the vehicle. If the difference between the two speeds lies below a tolerance or threshold value, it can be assumed that the traffic sign 103 is fixed in place. Experience has shown that fixed traffic signs of road users are usually to be considered. The recognized traffic sign 103 can thus be displayed to the driver of the vehicle and / or the speed limit signaled by the traffic sign can be used for automated vehicle control.

The recognition of the traffic sign recognized at the first time 103 can be simplified and / or accelerated if the search range can be narrowed down in a subsequent image. The vehicle speed is measured by a camera-independent speedometer of the vehicle at each time point and thus known. From the known vehicle speed and the known characteristics of the camera can thus be predicted a search range within which the traffic sign must be in a subsequent image. The prediction not only affects a change of location within the picture, but also a resizing. The predicted search area is in 2 exemplified by the dot-dashed frame 207 around the traffic sign 103 indicated.

3 shows a schematic representation of a camera view of a second traffic situation 300 in front of a vehicle at a first time. The traffic situation is similar to that of the 1 and 2 However, the road runs 102 in a left turn. To the road course of the road 102 to follow, the driver of the vehicle must turn the steering wheel to the left. Due to the steering wheel angle, the image captured by the camera changes accordingly; The change of direction corresponds to a camera pan. A traffic sign recognized in the picture taken at the first time 103 Thus, in a subsequent shot taken at a second time, a greater change in position than in FIG 1 and 2 described situation. If the speed of the camera and thus of the vehicle is determined solely by the change in size and position of the traffic sign in successive images whose time interval is known, the greater change in the position of the traffic sign due to the swiveling of the vehicle can lead to a less accurate determination of the speed lead or an incorrect speed is determined. This may result in a comparison of the speed determined by the camera images with the speed value supplied by the speedometer of the vehicle, the difference in the speeds being above a threshold, so that the traffic sign is not considered valid. A traffic sign not recognized as valid will not be displayed to the driver of the vehicle or will not be used for the automated vehicle control.

4 shows a schematic representation of the camera view of the second traffic situation 300 at a second time. As before with reference to 2 described the vehicle has moved forward by a certain distance and has come closer to the traffic sign invite three. Accordingly, the traffic sign 103 in the picture now larger and is located at a different position in the picture. The change of position compared to in 3 The camera view shown at the first time is indicated by the arrows 405 and 406 clarified. From the time difference between the first and the second time as well as the change of the position of the traffic sign 103 According to the invention, a speed is determined with which the camera has moved towards the traffic sign. According to the invention, in determining the speed of the Steering gear taken into account to the resulting greater change in the position of the traffic sign 103 to compensate. The greater change in the position of the traffic sign 103 opposite the comparable scene in 2 is clearly visible. As before with reference to the 1 and 2 described, the determined speed is compared with a velocity value supplied by a speedometer of the vehicle. If the difference of the two speeds lies below a tolerance value can be assumed that the traffic sign 103 is fixed in place. Experience has shown that stationary traffic signs are always to be considered by road users. The recognized traffic sign 103 can thus be displayed to the driver of the vehicle and / or the speed limit signaled by the traffic sign can be used for automated vehicle control.

This embodiment of the invention thus makes it possible to compensate for or reduce the influence of a lateral displacement of the scene captured by the camera when determining the speed.

The consideration of the steering wheel angle can also in the recognition of the recognized at the first time traffic sign 103 be used for simplification or acceleration by the search range is limited accordingly in a subsequent image. The search area is thus positioned not only from the expected change in the position of the traffic sign due to the approach of the vehicle to a traffic sign, but also due to the lateral displacement of the captured by the camera scene. In 4 is the search area as previously in 2 by a dot-dashed frame 407 characterized. The simplification or acceleration of the recognition of objects results in a shorter or lesser use of the resources provided in the vehicle for traffic sign recognition, which can be used accordingly for other purposes or put into a sleep mode. The prediction of the search range can also take into account other changes in the camera view due to driving dynamics, for example: B. a change in the inclination of the motor vehicle chassis in the longitudinal or transverse direction.

5 shows a schematic representation of a camera view on a third traffic situation 500 in front of a vehicle at a first time. As before in 1 shows the traffic situation straight to the horizon 101 running two-lane road 102 , The vehicle, not shown in the figure drives on the left lane in the direction of the horizon 101 and is about to get a truck 501 to overtake, which runs on the right lane with 80 km / h. At the back of the truck 501 is a sticker 502 attached, which represents a traffic sign. In the figure, the sticker shows 502 a traffic sign signaling a speed limit of 80 km / h and intended to indicate that the maximum speed of the truck 501 80 km / h. The road signs by the roadside 503 The maximum permissible speed indicated on this section of the road is 100 km / h. It is assumed that the traffic sign 503 has already been reliably detected and that the vehicle, not shown in the figure, is traveling at a speed of 100 km / h. The device of the invention recognizes that at the back of the truck 501 attached "traffic signs" 502 , and pursues it according to the invention on subsequently captured images of the camera.

6 shows a schematic representation of the camera view on the third traffic situation in front of the vehicle at a second time. It is clear that the traffic sign 503 disappeared from view of the camera while the truck 501 is in a slightly different position in the picture. In particular, the change in the size and position of the "traffic sign" 502 is compared to the previously recorded image only comparatively small. As before with reference to the 1 and 2 described the device according to the invention determined from the size change and the change in the position of the "traffic sign" 502 in successive pictures a speed. In the present example, the device according to the invention determines from the only slight change in the size and position of the "traffic sign". 502 a speed of 20 km / h, corresponding to the differential speed between the vehicle and the truck 501 , However, the speedometer of the vehicle provides a measured travel speed of 100 km / h. The difference of 80 km / h between the two speeds is above the threshold. It can be concluded from this that the traffic sign is not a valid stationary traffic sign. Accordingly, the incorrectly recognized traffic sign is not displayed or the information is not used for influencing driving parameters.

Also in this example, a search area in which the traffic sign is expected to be tracked in a subsequently captured image can be estimated from the speed of the motor vehicle. At a speed of 100 km / h, a valid, stationary traffic sign should be in the one indicated by the dot-dashed frame 507 highlighted area of the image will be recognized. The position of the frame 507 marks the expected search area. Already the fact that the "traffic sign" 502 has not been recognized in the expected range can be considered as an indication that it is not a valid, stationary traffic sign. At this point, the search for the recognition can be aborted and the signalization of the traffic sign to the driver or to an apparatus for automated driving omitted. Alternatively, the search range can be expanded or repositioned to reflect from an actual position of the "traffic sign". 502 in the picture to determine a speed. A difference between the speed determined from the image sequence and a speed determined by the speedometer of the motor vehicle is then compared with a threshold value in the manner described above. If the threshold is exceeded, ie the difference is too large, the signaling of the traffic sign to the driver or to an apparatus for automated driving is omitted.

Since the expected predicted search area for recognition and tracking of the traffic sign can already be determined using the data supplied by the speedometer of the motor vehicle before there is even an image to search, the search can be restricted immediately to the expected area. This makes the search faster and requires less resources or less resources for a shorter time. Further data from sensors of the motor vehicle, such as the steering wheel angle or an inclination of the vehicle, as described above, can be included in the prediction to improve the positioning of the search range.

7 shows an exemplary flowchart of an embodiment of the method according to the invention 700 , In step 702 First, a picture is taken in step 704 a pattern recognition for detecting traffic signs is supplied. If no traffic sign was detected, "no" branch of step 706 , the process is started from scratch. If a traffic sign was detected, "yes" branch of step 706 , another picture is taken, step 708 , the traffic sign in step 710 in the other image detected or tracked (tracking), and in step 712 determines a speed from the displacement and the size change of the tracked traffic sign in the images. The steps 708 . 710 and 712 may also be repeated one or more times (not shown in the figure), each time determining a speed from the sequence of images, and determining a mean velocity from multiple passes. Also in step 712 a speed value from the speedometer of the vehicle, either to each frame or after repeated execution of the steps 708 . 710 and 712 for a series of several pictures. In step 714 a difference is determined from the speed value related to the speedometer and the speed value determined from the tracking of traffic conditions over the series of images and compared to a threshold value. If the difference is above the threshold, "yes" branch of step 714 , the recognized traffic sign is discarded, ie not signaled to the driver or the device for automated driving. Otherwise, "no" branch of step 714 , the traffic sign is signaled to the driver or the device, which can then control the vehicle accordingly. It should be noted that the flowchart shown in the figure is only an example of a variety of design variants. For example, the capture of images for other purposes may be continuous, and only tracking may be conditional when a traffic sign has been detected. The recording of the additional image (s) can thus also consist in the conditional forwarding of images to a tracking module.

8th shows a schematic block diagram of a device 800 to carry out the process. A camera 802 is with a module 804 for traffic sign recognition, a tracking module 806 for tracking a recognized traffic sign and for determining a speed from the tracked traffic sign over a series of images, an interface module 808 , which obtains speed values from a speedometer of a motor vehicle, and a module 810 which determines a difference between the determined and the related speed and compares it with a threshold value, via one or more bus systems and / or data interfaces 812 connected. Optionally, in one or more embodiments of the device, a sensor module 814 be connected to other modules via the or the bus systems and / or data interfaces. The sensor module 814 can z. For example, values of vehicle sensors relating to a lateral or vertical inclination of the vehicle or a steering angle and the like, for example, to the module 804 for traffic sign recognition and the tracking module 806 to improve their respective function and / or accuracy. One or more of the modules may in this case be formed by respective computer programs which are executed on one or more processors in conjunction with correspondingly provided working memory means and non-volatile memory means. The bus system or the data interfaces 812 can be implemented in hardware or in software, so include software interfaces for data exchange between programs or program parts within a computer.

Claims (8)

Method for camera-based traffic sign recognition in a motor vehicle, comprising: - taking at least one image of a vehicle in front of a space with a camera, - performing a pattern recognition for detecting traffic signs in the at least one recorded image, - recording a series of other images of the front of the vehicle following the detection of a traffic sign in a first image, - tracking the traffic sign recognized in the first image in the sequence of the further images, - determining a first speed value indicating at which speed the detected and tracked traffic sign on the camera and / or moves past the camera, from a resizing and / or shifting of a position of the recognized and tracked traffic sign in the respective pictures via the sequence of the further pictures, - comparing the first speed value with one of a m speedometer of the motor vehicle provided second speed value representing a driving speed of the motor vehicle, the speedometer operates independently of the camera, and - Discard the detected traffic sign when the difference of the first and second speed values is above a threshold, or signaling the detected traffic sign a driver and / or a driver assistance system of the motor vehicle, characterized in that the method further comprises: - adjusting a frame rate, with which the sequence of further images is taken, or the threshold value in dependence on a steering angle and / or a change of an inclination of the vehicle Longitudinal and / or transverse direction. The method of claim 1, wherein the detected traffic sign is discarded if it could not be detected or tracked in an uninterrupted minimum number of further images. Method according to claim 2, wherein the minimum number is adjusted as a function of the second speed value, a steering angle and / or a change of an inclination of the vehicle in the longitudinal and / or transverse direction of the motor vehicle. The method of claim 1, wherein an estimate of a probable location of the tracked traffic sign in a subsequent image in dependence on the second speed value, a steering angle and / or a change of an inclination of the vehicle takes place in the longitudinal and / or transverse direction of the motor vehicle. The method of claim 1, wherein in the determination of the first speed value additionally a steering angle and / or a change of an inclination of the vehicle in the longitudinal and / or transverse direction of the motor vehicle are taken into account. The method of claim 1, wherein the method pauses at second speed values corresponding to a stationary vehicle. Contraption ( 800 ) for camera-based traffic sign recognition in a motor vehicle with a camera ( 802 ), a module for traffic sign recognition ( 804 ), a tracking module ( 806 ) for tracking a recognized traffic sign and for determining a first speed from the traffic sign tracked over a series of images, an interface module ( 808 ), which obtains speed values from a speedometer of a motor vehicle, a sensor module ( 814 ), which delivers values representing a transverse or longitudinal inclination of the motor vehicle and / or a steering angle to one or more further modules of the device and to a module ( 810 ), which determines a difference between the determined and the related speed and compares it with a threshold value, the modules being connected via one or more bus systems and / or data interfaces ( 812 ), the modules comprising computer programs running on one or more processors, which in their entirety cooperatively carry out the method according to one or more of claims 1 to 6. Data carriers with program code means, which are suitable for setting up on one or more arithmetic units of one or more control units of a motor vehicle to set up these for carrying out one or more of the method steps of claims 1 to 6.

Images